Csv 使用awk获取正确的列并定义要使用的分隔符和分隔符？

我有一个20000列的csv。这里是它的一个子集：

"eid","20216-2.0","20216-3.0","20220-2.0","20220-3.0"
"1548197","1","hello","","2020-03-05"
"2101984","2","string","","2020-03-04"
"2986696","3","no","","2020-04-05"
"1543304","3","ge","","2020-02-10"
"3207207","3","no","","2020-03-20"
"2373538","4","yesterday","","2020-03-01"
"4930973","5","today","","2020-03-06"
"6012673","54","tomorrow","","2020-05-05"
"4978627","1","yes","","2020-03-10"

我想使用awk获得两列：

awk -F "," '{ print $1, $3 }' input.csv  > output.csv

当我检查output.csv文件时，结果很混乱，如下所示：

"eid","20216-3.0"
"1548197","2020-03-05"
"2","string"
"no",""
"1543304",""
"","2020-03-20"
"yesterday",""
"4930973","2020-03-06"
"tomorrow","2020-05-05"
"4978627","2020-03-10"

有人能帮我吗

awk -v FPAT="([^,]+)|(\"[^\"]+\")" '{ print $1, $6713 }' input.csv  > output.csv

---

成功了！谢谢大家

怎么搞砸了？您展示的示例当然没有6713列，因此可能这就是问题所在，但我们确实无法告诉您预期的结果或如何解决此问题。我担心您确实需要澄清您现在得到了什么，以及您想要得到什么。另请参阅和，特别是提供a的指南。您的某些列是否可能在双引号中包含逗号？如上文所述，CSV文件中的字符串字段可以包含嵌入的引号，例如

“Smith，John”

或

“City，State”

。要正确处理这些问题，请尝试中的一种解决方案。Pandas知道如何处理带引号的字段，Awk只需按您的要求在每个逗号上拆分。不稳定行为的另一个可能来源是DOS回车，其中Python/Pandas更具弹性/Unix-y更少。但正如所贴，我认为这是不可复制的。你必须在你的FPAT中转义双引号，因为你在它周围使用了错误类型的引号。使用

FPAT='（[^，]+）|（“[^”]+”）“

。你也不需要paren，但它们可以提高清晰度，而且不会造成伤害。如果你有未引用的空字段，那么FPAT将失败，因此你最好在第一部分使用

*

，而不是

+

，如果你有嵌套的双引号，它将失败。总之，我建议你使用

FPAT='[^，]*.（“[^”]**）”+“

谢谢@EdMorton